Research On Topological MAP Based Bio-inspired Navigation

At present,inertial and satellite integrated navigation is the main means of navigation for unmanned combat platforms.When satellite signals are severely interfered or rejected,navigation system errors will rapidly accumulate over time and even cause the system to face the risk of “failure”.Therefore,high-precision auto-navigation has become one of the key technologies to be solved for unmanned combat platforms.Bionic navigation has become a hot spot in the field of navigation technology research.It is expected to provide a new technical solution for autonomous navigation of high-precision long-time navigation for unmanned combat platforms in complex environments.Taking the ground and aerial unmanned combat platforms as the background,this paper draws on the construction and recognition mechanism of the mammalian hippocampal region and the mechanism of the sensitive polarized light orientation of insect compound eyes,and focuses on the construction of navigation topology maps based on cell characteristics of cells.Multi-scale topological node identification and matching positioning method,multi-dimension polarization vision/inertial combination orientation method,and node-based recursive navigation algorithm based on topological map,and the technical solutions and bionics proposed by the on-board experiment and flight test data for the paper The feasibility of the navigation method was verified.The main research work and results of the paper are summarized as follows:(1)On the basis of an in-depth analysis of the activation characteristics and spatial expression structure of the grid cells in the mammalian hippocampus,the ground-air and airborne unmanned combat platforms are respectively presented,and a navigation topology map construction method based on cell characteristics of the grid is proposed;According to the motion state of the platform and the accuracy of the navigation system,the boundary constraints of the position and spatial scale of the topological node are given.Compared with existing navigation topology maps,the topology map constructed has a multi-scale,double-layered composite structure that can more effectively express and measure the motion space environment.(2)A multi-scale navigation topology node recognition algorithm is proposed.Aiming at the problem of weak distinguishability of scene features in Euclidean space,the spatial reconstruction algorithm for feature recognition based on LMNN is studied.The distinguishability of node feature distribution is enhanced,which is more conducive to recognition.Combined with ground and aerial unmanned combat platform The topological map of navigation,respectively,proposes a topological node recognition algorithm based on adaptive multi-scale and multi-scale image sequences,compared with the existing recognition algorithms,can significantly improve the correctness of node recognition;an improved node is given The feature matching localization algorithm reduces the number of the minimum matching points required to find unique solutions for PnP problems to two,which reduces the complexity of the algorithm and enhances the practicality.(3)The calibration and orientation algorithm of multi-polarized polarized visual heading sensor is studied.A calibration method based on LM for the multi-polarization polarization visual heading sensor is proposed,which can effectively improve the measurement accuracy of the sensor.A polarization image gradient noise reduction method based on polarization degree gradient(GDOP)is proposed to effectively suppress sky obstruction obstacles.Aiming at the influence of orientation accuracy,a global vision-based polarization multi-directional polarization vision/inertial combination orientation algorithm is proposed,and the method of solving the polarization directional ambiguity is given.The vehicle-borne experiment verifies the effectiveness of the algorithm.(4)A node recursive navigation algorithm based on topology graph is proposed.The algorithm is based on the navigation topology.According to the system accuracy of the system,the appropriate topology nodes are automatically established,the position observations obtained through the topological node identification and matching positioning,and the heading observations provided by the multi-polarization visual heading sensor are constrained.Conditions,the inertial information and the polarization heading information fusion,can effectively compensate for the cumulative error of the navigation system.The verification results of vehicle-mounted experiment and flight test data show that the algorithm can significantly improve the positioning and orientation accuracy of the navigation system.Even if the position observation information has a large “jump” during operation,the positioning and orientation error of the system still converges at a certain level.In scope,the effectiveness and usability of the method are demonstrated.